Using virtual machines (VMs) to perform hardware virtualization is becoming increasingly common. Typical storage systems store data as files, file systems, LUNs, and volumes, for example. When a typical storage system is used for storing and running virtual machines (VMs), the administrators who manage the storage systems must create and maintain a mapping between VMs and the typical storage abstractions. Mapping VMs to typical storage abstractions is inefficient in part because each VM generally requires multiple files, file systems, LUNs and/or volumes to store. The difficulty in mapping VMs to typical storage abstractions also makes imposing storage management policies at a desired granularity (e.g., on a per VM basis) inconvenient. Furthermore, typical storage systems also makes viewing statistics at a desired granularity (e.g., on a per VM basis) difficult.